Doubleword

What is Doubleword?

Low-cost async and batch inference API for open-weight models, pay per token.

Who is Doubleword best for?

Doubleword is best for teams with large LLM jobs that can wait hours, such as evals and data extraction.

• Large batch jobs

  Work that can wait about a day gets the lowest rates, up to 80% below the real-time price of the same model.

• Background agents

  Async requests come back in minutes to hours at up to half the real-time price, which suits agents that do not need instant replies.

• Evals and synthetic data

  Customers use it for dataset generation, annotation and evaluation runs, where token volume rather than speed drives the bill.

• Teams already on the OpenAI SDK

  The API is OpenAI-compatible, so switching mostly means changing the base URL and the model name.

How much does Doubleword cost?

Doubleword charges per token from prepaid credits with no minimum spend: both batch jobs (about 24 hours) and async jobs start at $0.02 per million input tokens, and each model has its own input and output rate.

Rates are set per model, with separate input and output prices, for about 30 open-weight models from DeepSeek, Qwen, GLM, Kimi, Gemma, Nemotron and others. Real-time requests cost full price and run on shared, rate-limited capacity meant for testing. At-Scale Inference adds dedicated real-time capacity, custom models, region-locked processing and volume pricing, and is priced on request. Self-service use is paid from prepaid credits, with no card needed to sign up. Full pricing on doubleword.ai. We checked these prices on 27 Sep 2026.

How do readers rate Doubleword?

No one has rated Doubleword yet. Be the first.

What are the best Doubleword alternatives?

One tool on Recaply does Doubleword's job.

Together AIPaid

Serverless and dedicated inference for open models, with fine-tuning.

Questions about Doubleword